A. Optical Edge Filters, Laser-Line Filters, and Their Use
Optical edge filters and laser-line filters are important components in systems for optical measurement and analysis including Raman spectroscopy and fluorescence spectroscopy. Optical edge filters and/or laser-line filters are used in such systems to block unwanted light that would otherwise constitute or generate spurious optical signals and swamp the signals to be detected and analyzed.
Optical edge filters block unwanted light having wavelengths above or, alternatively, below a chosen “transition” wavelength λT while transmitting light on the unblocked side of λT. Edge filters which transmit optical wavelengths longer than λT are called long-wave-pass filters (LWP filters), and edge filters which transmit wavelengths shorter than λT are short-wave-pass or SWP filters.
Referring to the drawings, FIGS. 1A and 1B schematically illustrate the spectral transmission of idealized long-wave-pass and short-wave-pass filters respectively. As can be seen from FIG. 1A, a LWP filter blocks light with wavelengths below λT and transmits light with wavelengths above λT. As shown in FIG. 1B, a SWP filter transmits light with wavelengths below λT and blocks light with wavelengths above λT. λT is the wavelength at which the filter “transitions” from blocking to transmission, or vice versa.
While an ideal edge filter has a precise transition wavelength λT represented by a vertical line at λT, real edge filters change from blocking to transmission over a small range of wavelengths and are more accurately represented by a non-vertical but steeply sloped line near λT. Similarly, while an ideal edge filter transmits all light in the transmission region (transmission T=1), real filters invariably block a small portion of the light to be transmitted (T<1). The steepness of the line and the proportion of the light transmitted are important parameters in many applications.
Turning now to FIGS. 1C and 1D, the spectral transmission of an ideal and a realistic laser-line filter (“LLF”) are illustrated respectively. Because lasers emit a very small, but non-zero, bandwidth of light, an ideal LLF passes light at wavelengths within this bandwidth (λL to (λL+BW)) with no ripple and perfectly steep (vertical) passband edges, as shown in FIG. 1C. The ideal LLF blocks light at wavelengths longer than the passband (λ>(λL+BW)) and blocks light at wavelengths shorter than the passband (λ<λL). The realistic LLF does not have complete transmission within the passband (λL to (λL+BW)) and has non-vertical passband edges, thereby changing from blocking to transmission over a small range of wavelengths, as shown in FIG. 1D. Accordingly, the steepness of the edges, the transmission, bandwidth, and the proportion of the light transmitted are important parameters of LLFs in many applications.
Edge filters and LLFs are particularly useful in optical measurement and analysis systems that use laser light to excite a sample at one wavelength λL (or a small band of wavelengths) and measure or view an optical response of the excited sample at other wavelengths. The excitation light λL is delivered to the sample by an excitation light path, and the optical response of the sample is delivered to the eye or measuring instrument by a collection path. Edge filters and LLFs can be used to block spurious light from the excitation path. Edge filters can be used to block excitation light from entry into the collection path. The steeper the filter edge(s), the more effectively spurious signals are blocked. In the case of edge filters, the lower the transmission loss, the more light from the sample reaches the measuring instrument. In the case of LLFs, the lower the transmission loss, the more excitation light from the laser reaches the sample.
Raman spectroscopy is one such optical analysis system. It is based on the fact that when molecular material is irradiated with high intensity light of a given wavelength (or series of wavelengths) λL, a small portion of the incident light scattered by the material will be shifted in wavelength above and below λL. This Raman shifting is attributed to the interaction of the light with resonant molecular structures within the material, and the spectral distribution of the Raman-shifted light provides a spectral “fingerprint” characteristic of the composition of the material. As a practical example, a Raman probe can identify the contents of a bottle without opening the bottle.
FIG. 2 is a simplified schematic diagram of a Raman probe 20. In essence, the probe 20 comprises an optical excitation path 22, and a collection path 23. These paths advantageously comprise optical fiber. FIG. 2 first will be described with the use of edge filters and then will be described with the use of a LLF.
In the case where edge filters are used, the Raman probe 20 may be configured to excite and collect the long wavelength portion of Raman-shifted light from a sample 21. In this case, edge filters 22A and 23A are disposed in the respective paths. In operation, excitation light λL from a laser 24 passes through the fiber path 22 and edge filter 22A to illuminate a portion of the sample 21 with high intensity light. Light scattered from the sample 21 passes through edge filter 23A and then through fiber collection path 23 to a spectral analyzer 25 where the “fingerprint” of the sample is determined.
Since a fiber path 22 through which the excitation signal passes is composed of molecular material, a small portion of the excitation light will be shifted in wavelength by the Raman effect in the fiber. This shifted light must be eliminated to prevent false readings. The removal can be accomplished by disposing a SWP edge filter 22A between the fiber 22 and sample 21. SWP edge filter 22A, having a transition wavelength just above the laser wavelength λL, blocks both long wavelength Raman scattering from the fiber and long wavelength noise from the laser.
However, in some cases, a filter 22A is not needed. For instance, if the excitation signal is direct laser radiation that does not pass through molecular material at any appreciable length, and contains no appreciable light at wavelengths other than the desired laser line itself, then the filter 22A is not necessary. An example would be direct laser radiation illuminating a sample 21 through a vacuum.
The light scattered from the sample 21 is a mixture of unshifted scattered excitation light λL (Rayleigh scattering) and Raman-shifted light at wavelengths longer and shorter than λL. The scattered excitation light would not only swamp the analyzer, it would also excite spurious Raman scattering in a collection fiber. Thus the unshifted excitation light should be removed from the collection path. This can be accomplished by disposing a long pass edge filter 23A between the sample 21 and the collection fiber 23, the long pass filter having a transition wavelength λT just below the excitation wavelength λ. This arrangement ensures that the light reaching the analyzer is predominantly the long wavelength Raman-shifted light from the sample. Analogous arrangements using edge filters can be used to analyze short wavelength Raman-shifted light.
FIG. 2 also may use a LLF to excite Raman-shifted light from the sample 21. In this case, a LLF 22A and a blocking edge or notch filter 23A are disposed in the respective paths. In operation, excitation light from a laser 24 passes through the fiber path 22 and the LLF 22A to illuminate a portion of the sample 21 with high intensity light of a wavelength (or series of wavelengths) λL. Light scattered from the sample 21 passes through the blocking edge or notch filter 23A to block λL and then through fiber collection path 23 to a spectral analyzer 25 where the “fingerprint” of the sample is determined.
As with the edge filter configuration, since the fiber 22 through which the excitation signal passes is composed of molecular material, a small portion of the excitation light will be shifted in wavelength by the Raman effect in the fiber. This shifted light is eliminated by LLF 22A. LLF 22A passes the laser wavelength and blocks both long and short wavelength Raman scattering from the fiber.
The unshifted scattered excitation light is removed from the collection path. The notch or edge filter removes the laser light of wavelength λL. This arrangement ensures that the light reaching the analyzer is predominantly Raman-shifted light from the sample.
Edge filters and LLFs are also useful in fluorescence spectroscopy. Here, laser excitation light is used to excite longer wavelength emission from fluorescent markers. The markers can be fluorescent atoms chemically bonded to a biological molecule to track the molecule in a body or cell. LLFs may be used, as in Raman spectroscopy, to reject spurious light from the excitation path. Notch filters may be used to reject excitation light from the collection path. Edge filters may be used to reject spurious light from the excitation path and to reject excitation light from the collection path.
In the case of edge filters, it should now be clear that the steeper the filter slope at the transition wavelength λT the greater the amount of spurious light that can be filtered out. In addition, the steeper the slope, the greater the amount of shifted light from the sample that will reach the analyzer. Similarly, higher levels of transmission of the shifted light through the filters provide more light for analysis. Higher edge filter blocking provides better rejection of the laser excitation light from the spectrum analyzer, thus decreasing the noise and improving both specificity and sensitivity of the measurement. Higher edge-filter transmission enables the maximum signal to reach the analyzer, further improving the signal-to-noise ratio and hence the measurement or image fidelity. A steeper filter edge also permits shifts to be resolved much closer to the excitation wavelength, thus increasing the amount of information from the measurement.
In the case of LLFs, the steeper the edges of the LLF at the laser wavelength λL, the greater the amount of spurious light that can be filtered out. Similarly, higher levels of transmission of the laser light through the LLF, the more light for excitation.
B. Edge Filter and LLF Structure and Conventional Fabrication
FIG. 3A is a simplified schematic illustration of an optical edge filter 30. The optical edge filter 30 comprises a transparent substrate 31 having a flat major surface 32 supporting many thin coatings 33A, 33B. The thickness of the coatings is exaggerated and the number is reduced for purposes of illustration. Coatings 33A and 33B are typically alternating and of different respective materials chosen to present markedly different indices of refraction (index contrast). The coating indices and thicknesses are chosen and dimensioned to filter impinging light by interference effects in a desired manner. Specifically, if a light beam 34 impinges on the filter, a first wavelength portion 34T of a beam is transmitted and a second wavelength portion 34R is reflected and thus rejected by the filter. What is transmitted and what is reflected depends on the precise thicknesses and indices of the thin coatings.
Two basic types of thin-film edge filters and thin-film laser-line filters exist: those based on “soft coatings” and those based on “hard coatings,” both of which are typically manufactured by an evaporation technique (either thermal evaporation or electron-beam evaporation). Hard coating filters, however, may also be manufactured by non-evaporative techniques such as ion-beam sputtering.
Soft coatings imply literally what the name suggests—they are physically soft and can be readily scratched or damaged. They are fairly porous, which also means they tend to be hygroscopic (absorb water vapor) leading to dynamic changes in the film index and hence the resulting filter spectrum in correlation to local humidity. There are two main reasons soft coatings are used. First, an advantageous larger index contrast can be realized with soft coatings. (The index contrast is the relative difference between the index of refraction of the low-index material and that of the high-index material.) For example, many high-performance soft-coated filters are made using sodium aluminum fluoride (“cryolite”), with a chemical composition of Na3AlF6 and an index of about 1.35 for visible wavelengths, and zinc sulfide, with a chemical composition of ZnS and an index of about 2.35. The second reason for using these materials is that the evaporation process can be controlled well for these materials, largely because they have relatively low melting temperatures. Hence it is possible to maintain fairly accurate control over the layer thicknesses even for filter structures with many 10's of layers and perhaps even up to 100 layers. As described above, edge filter performance is measured by edge steepness, depth of blocking, and high transmission with low ripple. A larger index contrast and a larger number of layers both yield more steepness and more blocking. High transmission with low ripple is improved with more layers and higher layer thickness accuracy. For these reasons the highest performance conventional thin-film edge filters have been made with soft-coating technology.
Hard coatings are made with tougher materials (generally oxides), and result from “energetic” deposition processes, in which energy is explicitly supplied to the film itself during the deposition process. This is accomplished with a beam of ions impinging directly on the coating surface. The ion bombardment acts to “hammer” the atoms into place in a more dense, less porous film structure. Such processes are usually called ion-assisted deposition (IAD) processes. High-performance edge filters have been made with ion-assisted electron-beam evaporation. Typically the index contrast available with hard-coating (oxide) thin-film materials is not as high as that of the soft-coating materials, and consequently more layers must be deposited to achieve a comparable level of performance. This problem, coupled with the more difficult to control deposition rates and overall processes of high-melting-temperature oxides, leads to much more stringent requirements on the layer-thickness control techniques to achieve a reasonable level of layer thickness accuracy for good edge steepness and high, low-ripple transmission.
For the best filters, some kind of “optical monitoring” (direct measurement of filter transmission or reflection during deposition) is necessary to determine when to terminate the deposition of each layer. Optical monitoring can be performed on the actual filters of interest or on “witness pieces” often positioned in the center of the deposition chamber. There are three basic types of optical monitoring algorithms. The first is often called “drop-chip” monitoring, and is based on measuring the transmission (or reflection) vs. time through a new witness piece for each new layer. Since the theoretical transmission vs. time can be calculated accurately for each layer deposited on a blank piece of glass, then a good comparison between the measured and theory curves can be made independent of the history of the deposition (thickness errors in previous layers). This technique is accurate and useful for layers of arbitrary thickness, but it is cumbersome, especially for filters comprised of at least many 10's of layers.
The second type of monitoring is called “turning-point” monitoring, and is used for depositing layers that are precisely a quarter of a wavelength in thickness (or multiples thereof). The technique is based on the fact that the transmission vs. time reaches a turning point (or extremum) at each multiple of a quarter wave of thickness, so an algorithm is developed to cut layers precisely at the turning points. The elegant feature of this method is that there is inherent compensation for layer thickness errors from previous layers, so long as one adheres to the rule of cutting exactly at turning points. It thus works extremely well even for very thick coatings with even hundreds layers (it is the basis for manufacturing very high-performance filters for DWDM telecom applications, which can have as many as 200–400 quarter-wave layers).
The third type of monitoring is called “level monitoring,” and is applicable for non-quarter-wave thick layers. Monitoring can be done through the actual filters or through witness piece(s). The concept is to cut layers at predetermined transmission levels, based on a calculated prediction of transmission vs. time for the entire structure. However, because small layer errors lead to large variations in the absolute transmission values, one must instead rely on cutting at the correct transmission level relative to the local maximum and minimum values. Hence the method works well only for non-quarter-wave thick layers that are more than a half-wave thick, so that there is both a maximum and a minimum transmission value in the transmission vs. time curve for that layer. Even in this case, this method does not contain inherent compensation for errors in the thickness of previously deposited layers, and thus is not as forgiving as the turning-point method. However, to obtain an edge filter with high transmission and low ripple requires primarily non-quarter-wave thick layers, and hence turning-point monitoring is not applicable for edge filters.
Turning now to FIG. 3B, a typical conventional laser-line filter 300 of the thin film type is illustrated. The filter 300 comprises a bandpass section 301 to define the passband and provide blocking in the immediate spectral vicinity of the passband. It further comprises a blocker section 302 to provide additional blocking at wavelengths further away from the passband.
The bandpass section 301 comprises a sequence of dielectric layers forming a sequence of Fabry-Perot cavities combined with appropriate “coupling layer” spacers between them. Each Fabry-Perot cavity comprises alternating layers of high index material 303 and low index material 304, where most layers have a thickness of one quarter wavelength at the passband center wavelength. In isolation, each quarter wave stack would provide high reflection blocking of a broad stopband about the center wavelength. However, when a pair of such quarter wave stacks are coupled to one another with an intervening half-wavelength cavity layer 305 of high or low index material, a Fabry-Perot cavity is formed, and a narrow, high-transmission passband is opened in the center of the stopband region. When multiple Fabry-Perot cavities are combined with appropriate “coupling layers” between them, the narrow passband region achieves a more rectangular passband shape, with steeper edges and a proportionally wider, flatter region of high transmission. While most laser-line filters are based on at least two Fabry-Perot cavities, it is not uncommon for laser-line filters to be comprised of three or more cavities. The stopband associated with a multi-cavity Fabry-Perot structure blocks light over only a limited region of wavelengths. For instance, for a structure based on SiO2 as the low-index material (n=1.5) and Nb2O5 as the high-index material (n=2.3), the stopband ranges from about λc/1.14 to λC/0.86, where λC is the center wavelength of the passband.
To provide additional blocking at wavelengths further away from the passband, prior-art thin-film filters rely on the addition of the blocker section 302 comprising a combination metal-dielectric blocking coating 306, or absorbing colored glass 307, or both. The metal-dielectric coating can be designed to have a passband centered on the narrow passband associated with the multi-cavity Fabry-Perot coating, but the metal-dielectric coating passband is generally very broad. Absorbing colored glass is used primarily to block short wavelengths. The major disadvantage of such additional coatings and colored glass is that they severely limit the transmission of light in the passband due to optical absorption losses in the metal layer(s) and/or the absorbing glass. Therefore, the passband transmission for such filters typically does not exceed about 70% and often is as low as 10% (for most filters the transmission is in the range 20–50%).
Holographic laser-line filters are comprised of a volume diffraction grating, a mount, and some kind of spatial filtering apparatus. It functions as follows: first, a collimated, linearly polarized laser beam is directed perpendicularly into the mount that holds the volume diffraction grating (which itself is comprised of the holographic material, often dichromated gelatin, sandwiched between pieces of glass to protect it from environmental and physical contact). The internal transmission grating diffracts the laser-line through 90 degrees so that the output beam is nominally perpendicular to the input laser beam and the exit face of the mount. The grating is spectrally selective so that only the light which is very close to the design wavelength of the filter is diffracted. The undiffracted light passes through the back face of the mount. Unwanted light with wavelengths near the laser-line is diffracted by the transmission grating but at different angles than the laser-line. This unwanted diffracted light is then removed by a spatial filter (slit or aperture) located beyond the exit face of the cube. The overall apparatus provides effective filtering, but is very inconvenient to use due to the redirection of the laser beam, the bulky nature of the setup, and the need to carefully align the spatial filter with the mounted diffraction grating. Furthermore, the requirement that the desired laser beam be linearly polarized in a particular orientation is severely limiting to the applicability of this filter.
Accordingly there is a need for an improved method of making optical edge filters and LLFs and for improved edge filters and LLFs having increased edge steepness and increased transmission.